Post-trade Blockchain – no way out of this maze

A rat in a maze is free to go anywhere, as long as it stays inside the maze. Margaret Attwood

In future, September 2017 may be seen as the inflection point when the debate on blockchain changed from the religious to the scientific. Two important reports were produced which poured very cold water on some of the wilder claims made for the benefits of blockchain, aka as distributed ledger technology (DLT).

In the first report, the Bank of International Settlements (BIS) published a detailed analysis of ‘central bank cryptocurrencies’ by two respected academic economists, who are also experts in payments systems. After classifying and discussing the histories and potential futures of various types of currency, including different cryptocurrencies such as Bitcoin, the authors concluded that unless anonymity is a prerequisite (other than for money-laundering, of course) satisfactory solutions already exist.

The second paper by the European Central Bank is a comprehensive analysis of the “potential impact of DLTs on securities post-trading harmonization and on the wider EU financial market integration”. It is very serious piece of work, conducted by experts in post-trade settlement systems, which should be studied by all of those interested in the potential uses of DLT.

The ECB ‘advisory group on market infrastructures for securities and collateral’ pays DLT the compliment of making the assumption that DLT is a potential solution to a real problem, and then works to tease out the real-life implications and limitations of using such a technology. It is a significant piece of real analysis, rather than a grandiose statement of potential futures. The paper attempts to answer the question - what would a viable DLT solution look like for post-trade settlement (which has often been considered the ideal use-case for DLT)?

In 134 densely filled pages, the paper describes the analyses undertaken by the ECB team and the various scenarios that were considered, in some detail.

Unfortunately, the results of the various analyses resemble the path of a mouse in a maze[1].

In the study, the researchers would suggest a scenario for implementation of DLT and then carefully consider the implications and consequences of that scenario. And like the aforementioned rodent they would go down each sub-branch of the maze before finally giving up and going back to try another branch. There was, unfortunately for the researchers, no Eureka or lightbulb moment - there was no exit!

The researchers very quickly reversed out of and dismissed one branch of the maze – a ‘public’ blockchain/DLT with “transparency[2]” and “Immutability[3]”

“As long as illicit behaviour can be detected by harmed parties and disputes resolved by proper governance or judicial frameworks, the necessity for public ledgers disappears, allowing for some level of data confidentiality (see Chapter 13[4]), although this may mean foregoing part of the cyber resilience brought by data replication in DLT networks (see Chapter 11)”

In effect, this means that the potential solutions boil down to a comparison between different configurations of a ‘permissioned’ DLT (with access controlled and monitored by some ‘centralized authority’) and the existing architectures of a central securities depository (CSD), such as the DTCC or the ASX, in a securities settlement system (SSS[5]).

The paper provides some must-read sections that define and describe current settlement processes (which are worth having in every library for reference purposes) but then quickly turns to the biggest branch in the maze – DvP or Delivery versus Payment, which is the raison d’etre of settlement systems:

“DvP settlement links a transfer of securities with a transfer of cash in a way that the delivery of securities occurs if and only if the corresponding transfer of cash occurs and vice-versa. This is generally seen as an effective method of addressing principal risk, as it avoids a situation where participants in a securities transaction transfer the asset they committed to a trade (securities or cash) before the asset they expect to receive (cash or securities) is guaranteed to be transferred in the opposite direction”.

In other words, both delivery AND payment must take place or neither does. Note they do not necessarily have to occur simultaneously[6] merely that both ‘legs’ are guaranteed (usually by a CSD).

The report notes that DvP is ‘straightforward’ and can be guaranteed in a DLT environment

“when the securities leg and the cash leg are settled in the same ledger and are governed by the same DLT protocol, or when settlement in two different ledgers can be linked by means of automatic escrow services”

This of course is a restatement of the known property that DLT only works in a ‘closed’ system, here one that handles both securities AND cash, OR processes are handled by another third party (here an ‘escrow service’). But, of course, in current settlement systems, cash is handled, for very good reasons of efficiency[7], through the banking system and managed by the CSD, and ‘escrow’ is provided by a CSD. So, this scenario is not removing the need for a CSD-like entity, merely replacing a part of it.

Having backed out of the ‘all or nothing’ branch of the maze, the report dives into some other even less attractive branches for DvP. The first is the use of ‘central bank money’ (CeBM)

“If a DLT solution were to be adopted for the bookkeeping of securities, the advantages that DvP settlement in CeBM brings in terms of risk mitigation might justify also reflections over making CeBM available on a distributed ledger. This could be useful if it is assumed that no interfacing between the DLT securities settlement system and current non-DLT CeBM payment systems will be able to yield the same safety and efficiency gains”

However, that exit route is closed almost immediately as the report noted “the ECB cannot, at this stage, consider basing [its] market infrastructure on a DLT solution”. In other words, the central bank is not going to change the banking system just to accommodate securities settlement (which happens to work quite adequately at present).

The report then discusses the legal difference between “transfer of ownership”, which is native to DLT solutions, and “settlement finality”, which is not. Settlement finality occurs to cover the situation when a participant becomes insolvent. In the known use-case of DLT, i.e. Bitcoin, such a situation can happen if goods paid for by BTC do not get delivered, and there is no recourse. The research then hits another dead end

“As stated above, settlement finality needs to be considered on each settlement system level separately. This would render irrelevant the distinction between, for instance, whether the assets were “digitised/tokenised” or “digital/native” assets. For example, it would not make any difference to the settlement finality from the investor CSD’s point of view whether the issuer CSD was using a DLT-based settlement system or the current T2S platform. Settlement finality would need to be defined in a DLT-based settlement system as well’

In other words, DLT “tokens” are irrelevant to real-life issues of investor protection.

The paper then raises issues of timing which in a decentralized model becomes more important than in a centralized model (which is the timing authority) stating “the decentralised nature of a DLT solution creates certain technical and/or operational challenges when it comes to ensuring that single moments of finality are defined”. That problem too is left hanging, unsolved.

Having encountered dead-ends everywhere the researchers go down another “all or nothing” path, by considering the expansion of a DLT system to cover full trading activities, i.e. pre- and well as post-trade processes.

However, the researchers realize very quickly that they are entering into an even bigger maze and back out pretty quickly recognizing that a lot of things have to happen before a DLT can be considered for trading and that the current systems would still be needed

“However, there will still be a need for a comprehensive SDR [Settlement Discipline Regime] if DLT-enabled processes do not fully contain all the following characteristics:

1)     no difference between a trading position and a settlement position, e.g. a security can only be sold from a settled position, or securities fail lending is integrated in such a way as to ensure settlement;

2)     the largest possible universe of market participants involved in trading in a specific asset (e.g. ISIN) is represented on the blockchain, thus removing the need to reconcile with databases outside the system that otherwise could cause further operational errors;

3)     no operational dependencies on systems that do not interoperate with the DLT system, e.g. settlement of the cash leg must be possible either via DLT or in a way that does not affect the consistency of data accessible by the DLT system in real time;

4)     the instruction method is highly automated, with unambiguous static data to ensure low possibility for human error to instruct trades;

5)     assets in securities financing transactions are fully fungible, e.g. if an asset is subject to a collateral arrangements [sic] there is no risk of assets being recalled when needed as part of a substitution process.

Of course, this set of conditions is merely a restatement of the “all or nothing model” of DLT and ignores the realities of financial trading, not least that trading and settlement positions are very different. In short, a DLT solution won’t work here either!

The report then goes on (in excellent challenges and opportunities sections) to discuss several other important issues such as ‘settlement risk’, ‘privacy’, ‘interoperability’, ‘standardisation’, ‘taxation’, ‘multiple regions’ and so on, which may be summed up by one comment about potential hurdles

“Unless all segments of financial markets adopt the same schedules, there may be significant hurdles arising at the interconnecting points (e.g. significant funding and transferability issues between DLT-based settlements and traditional settlements).”

Unfortunately, there is no exit to this maze, as each new ‘opportunity’ throws up new, and arguably greater, 'challenges'. Maybe the current centralized approach, though creaking at the edges, might not be a bad alternative after all. DLT is no silver bullet.

The report does not go into implementation risks in detail but does make the statement

“Although the possible benefits are quite evident[8] and very appealing on a theoretical basis, the complexity of how to actually achieve the transition to such a new DLT-based framework should not be underestimated, as very elaborate coordination of technical developments, business incentives and conversion timetables would need to be established and centrally managed in order to ensure a seamless, concurrent transition to the new framework by all actors”.

In other words, the chances of getting everyone on board to implement such an industry-wide initiative are not great, in fact the chances are very low.

The report concludes with

“It is important to note that a number of elements of a theoretically DLT-enabled financial market have to be properly designed and put together before DLT adoption can be considered a realistic possibility in the securities settlement space”.

In short, DLT is not a viable solution yet for post-trade settlement. And while the report recommends that current standards be strengthened it does leave open the door to future innovations [as good researchers should]. But given this extensive analysis, the author suggests that such innovations are unlikely to be in found the DLT maze.

It would be a very brave (or foolhardy) organization that embarked on an industry-wide conversion to DLT technology today without covering off ALL of the issues and risks raised by this excellent report.

[1] Before moving on, the author freely admits that he is on-record as being highly sceptical concerning the potential uses of DLT in finance and is open to the potential for cognitive biases, not least confirmation bias. However, aware of that, the ECB paper deserves serious analysis.

[2] The researchers dismiss the often-stated ‘requirements’ for transparency - in fact market participants don’t want transparency because “the confidentiality of data among market participants is necessary inter alia to hide trading strategies, limit price volatility, and protect the privacy of end investors”.

[3] The paper dismisses the requirement for “immutability’ referring instead to the concept of ‘finality’, “[…] public authorities, financial institutions and market infrastructures are aware that different levels of finality can be attached at different legally defined moments. Finality is defined by the system operator under the applicable national law”.

[4] Note the chapter numbers have been retained in quotations to allow readers to access the paper directly.

[5] The researchers identify the subtle legal differences between a SSS and a CSD and conclude that in order to satisfy European law, a DLT that operates as a DLT sttlement system (SSS) "would need to be operated by a CSD that complies with CSDR [Central Securities Depository Regulation] requirements (see Article 18(2) of the CSDR), including minimum capital requirements, conduct of business rules, prudential rules, etc.” That means that some form of CSD is required, unless of course all of the securities and consumer laws were to be changed. 

[6] For example, multiple payments may be netted before being cleared.

[7] Such as netting payments in multiple buy/sell transactions.

[8] The author would argue that as the report did not do any cost-benefit analysis that the benefits are far from ‘evident’.